Using NVFAIL

How the NVFAIL option provides database protection

The nvfail option of the vol options command provides database file protection by detecting NVRAM inconsistencies at cluster failover or at boot time.

If this option is enabled and problems are found, it warns database administrators of NVRAM problems by

Stopping response or shutting down database instances
Sending error messages to the console alerting you to check the state of the database

How you can provide additional protection for databases

You can provide additional protection to specific database LUNs by adding them to an optional file you create called /etc/nvfail_rename. When you enable nvfail and Data ONTAP detects NVRAM errors at bootup, Data ONTAP renames any database LUNs specified in the nvfail_rename file by appending .nvfail to the original name. When Data ONTAP renames a database LUN, the database cannot restart automatically. This gives you the opportunity to examine it for inconsistencies before you remove the .nvfail extension and make it accessible again.

How NVFAIL works in the SAN environment on the filer

LUNs accessed concurrently over NFS and FCP can be entered in /etc/nvfail_rename. This causes a LUN rename upon boot after an NVRAM failure. This process is described in more detail in the NVRAM chapter of the Data ONTAP 6.4 System Administration Data Protection Guide.

Caution
NVRAM failure can indicate data corruption. Upon a reboot after an NVRAM failure, all LUNs are disabled and are no longer exported over FCP. You need to reenable (and reexport) each LUN manually to the initiator after data recovery.

Enabling the NVFAIL option

To enable the NVFAIL option on WAFL volumes, complete the following step.

Step	Action
1	Enter the following command: *vol options volume-name* on**

Creating the nvfail_rename file

To create the nvfail_rename file, complete the following steps.

Step	Action
1	Use an editor to create or modify the nvfail_rename file in the filer's /etc directory.
2	List the path name and file name, one file per line, within the nvfail_rename file. Example /vol/vol1/home/dbs/oracle-WG73.dbf
3	Save the file.

Step

Action

Use an editor to create or modify the nvfail_rename file in the filer's /etc directory.

List the path name and file name, one file per line, within the nvfail_rename file.

Example

/vol/vol1/home/dbs/oracle-WG73.dbf

Save the file.

How to check for NVFAIL on the Solaris SAN host

You can use the sanlun lun show command on the host to show whether there was an NVRAM failure involving one or all LUNs at a given path name. The output of this command is preceded by a line of column headings, the last of which, lun state, can be GOOD, OFFLINE, NVFAIL or ERROR.

Example

# sanlun lun show 

filer: lun-pathname device filename adapter lun size lun state 

filerA: /vol/vol0/lun0 /dev/rdsk/c1t1d0s2 lpfc0 2.0g (2147483136) 
GOOD 

filerA: /vol/vol0/lun1 /dev/rdsk/c2t1d0s2 lpfc1 2.0g (2147483136) 
GOOD 

filerA: /vol/vol0/lun2 /dev/rdsk/c2t1d0s2 lpfc1 2.0g (2147483136) 
NVFAIL